What is a Quasi-Experimental Design?

Let’s explore the most common types of quasi-experimental designs:

• Regression Discontinuity

The regression discontinuity approach involves measuring the impact of the treatment, or independent variable, by applying a treatment assignment mechanism that is based on a continuous eligibility index that is variable with a continuous distribution.

Subjects are therefore selected to be a part of the treatment group based on whether their value on a predetermined numeric rating scale exceeds a designated threshold. Those who fall below that threshold, do not receive the treatment and will hence be part of the control group instead. 

• Non-equivalent Groups Design

In the non-equivalent groups design, researchers select two groups; one is exposed to the treatment while the other is not. This is not considered random assignment as the researcher is working with pre-existing groups and not allocating subjects to either group based on random assignment. Although researchers aim to select two groups that are as similar as possible, we can never be sure that the groups are actually comparable. It is highly unlikely that the two groups would be as comparable as they would have been if they were created using random assignment. This design was named as the nonequivalent groups design for this very reason, as it’s likely that the two groups are not equivalent. 

• Natural Experiments

In both laboratory and field experiments, researchers generally determine how subjects are assigned to groups, whether that be through random or non-random criteria. However, in natural experiments, a naturally occurring external event or situation results in the random assignment of subjects to different groups. Therefore the factors that influence assignment are out of the control of investigators.

Although many natural experiments employ a method that resembles random assignment, they are still not considered to be true experiments because they utilize the observational method. Some situations in which natural experiments are appropriate and commonly used include policy changes, weather events, and natural disasters.